Drawing Operations

Drawing Together

In a large, brightly lit performance space, Sougwen Chung kneels over a massive sheet of paper, brush in hand. Beside her, a robotic arm — sleek, articulated, mounted on a low platform — holds its own brush. Chung makes a mark. The robot watches, processes, and responds with a mark of its own. Chung adjusts. The robot adjusts. Over minutes and hours, a drawing emerges that neither human nor machine could have produced alone.

This is Drawing Operations, a series of performances and installations that Chung has been developing and refining since 2015. The project has evolved through multiple iterations — from early experiments where a single robotic arm mirrored Chung’s gestures, to complex multi-robot systems that respond not only to the artist’s movements but to environmental data, audience presence, and the evolving drawing itself.

The result is not a human drawing with robotic assistance, nor a robot drawing with human supervision. It is a genuine collaboration — a visual conversation between two radically different forms of intelligence, unfolding in real time before an audience.

The Technology

The robotic systems in Drawing Operations are custom-built by Chung and her studio. The robots’ movements are driven by recurrent neural networks — a type of machine learning model particularly suited to sequential data — trained on thousands of Chung’s own drawings. The models have learned the patterns, rhythms, and tendencies of Chung’s hand: how she begins a stroke, how she varies pressure, where she tends to place marks in relation to existing lines.

But the robots do not simply copy Chung. They interpret her movements through the lens of their training, producing responses that are recognizably related to her style but distinct from what she would draw herself. The effect is something like watching a musician jam with a partner who knows their tendencies well enough to anticipate and complement them, but who has their own instincts and surprises.

In more recent iterations, Chung has added environmental sensors that feed additional data into the robots’ decision-making. Temperature, humidity, sound levels, and the movement of audience members all influence the robotic marks. This means that every performance is unique not only because of the improvised interaction between Chung and the machines but because the environment itself participates in the creation.

The Artist

Chung’s path to Drawing Operations reflects a career-long engagement with the relationship between human and machine creativity. Born in Canada and raised in Hong Kong, she studied at the MIT Media Lab and was a research fellow at Bell Labs, bringing a deep technical background to her artistic practice. She was a former research fellow at the New Museum’s NEW INC incubator and has held positions at institutions spanning art, technology, and design.

This dual fluency — in both the art world and the technology world — is essential to what makes Drawing Operations work. Chung is not an artist dabbling in technology or a technologist dabbling in art. She operates with genuine expertise in both domains, and the work reflects that integration. The robotic systems are technically sophisticated, and the artistic choices — scale, gesture, composition, performance structure — are informed by deep knowledge of drawing history and contemporary art practice.

Chung has spoken extensively about the emotional experience of drawing alongside machines. She describes the early sessions as unsettling — the robots’ movements were eerily familiar, echoing her own gestures in ways that felt like looking in a distorted mirror. Over time, the relationship evolved into something she describes as a genuine partnership, with the robots’ unexpected responses pushing her in creative directions she would not have explored alone.

The Performances

Drawing Operations performances are mesmerizing to watch. There is an intimacy to the interaction between Chung and the robots that is immediately legible even to audiences unfamiliar with the underlying technology. The human artist moves with the fluid, intuitive quality of someone deeply connected to their practice. The robot responds with a different kind of fluidity — precise but slightly alien, following its own logic.

The drawings that emerge are large-scale, abstract, and layered. Human and machine marks interweave, sometimes harmonizing, sometimes creating visual tension. You can often distinguish Chung’s marks from the robots’ — hers carry the subtle irregularities of a human hand, while the machines produce strokes that are smoother and more geometrically consistent — but the overall composition is inseparable from the collaboration. Neither set of marks would make sense without the other.

Some performances are solo affairs — Chung and one or two robotic arms in a quiet studio. Others are large-scale public events with multiple robots, live music, and hundreds of spectators. The scale and format vary, but the core proposition remains consistent: what happens when a human and a machine create together, in real time, with neither fully in control?

Why It Matters

Drawing Operations matters because it offers a model of human-AI collaboration that avoids the extremes that dominate the conversation. It is not AI replacing an artist. It is not an artist using AI as a passive tool. It is a genuine partnership — messy, unpredictable, and productive in ways that neither party could achieve independently.

Chung’s insistence on training the robots exclusively on her own work addresses the consent problem that plagues much of AI art. The machines learned from her drawings, with her knowledge and active participation. This is not a minor detail — it transforms the ethical character of the entire project.

The physicality of the work is equally significant. In a discourse dominated by screen-based images, Drawing Operations produces real objects — large drawings on paper, marked by real brushes loaded with real ink. You can see the texture of the paper, the pooling of ink where a stroke began, the slight smudging where a robotic arm moved through a still-wet line. The work has presence in a way that a digital image cannot match.

Drawing Operations suggests that the most artistically rich applications of AI may not be in generating finished products but in creating new kinds of creative relationships — between humans and machines, between intention and surprise, between control and emergence. It is not a vision of the future in which AI makes art for us. It is a vision in which AI makes art with us.